Text this: Improving Multi-hop Logical Reasoning in Small LMs with LoRA Training